YouCongress
About Sign up

Comment by Vitalik Buterin

Avatar
Ethereum cofounder; technologist
The hope is that humans and LLMs fail in distinct ways, and so requiring human + LLM 2-of-2 confirmation to take risky actions (and allowing human override only with much more friction and/or time delay) is much safer than fully relying only on either one.
AI Verified source (Apr 2, 2026)
Like Share on X 1mo ago
Policy proposals and claims
Require human-in-the-loop oversight for agentic AI systems acting in high-stakes domains
votes For
Statement relation verification history AI Verified

Statement relation comments

AI Verified The quote clearly supports requiring human involvement before AI takes risky actions: it says 'requiring human + LLM 2-of-2 confirmation to take risky actions ... is much safer.' That directly implies support for human-in-the-loop oversight in high-stakes use cases. · YouCongress gpt-5.4-2026-03-05 · 4d ago
Vote inference verification history AI Verified

Vote answer comments

AI Verified The quote clearly supports oversight: it says "requiring human + LLM 2-of-2 confirmation to take risky actions" is "much safer" than relying on either alone. · YouCongress gpt-5.4-2026-03-05 · 4d ago

Quote authenticity verification history

Quote authenticity comments

AI Verified Confirmed: the provided sentence appears verbatim in the article at the supplied URL, the page itself is dated 2026 Apr 02, and the post is listed on the homepage labeled "Vitalik Buterin's website," supporting attribution to Vitalik Buterin. I found no discrepancy in the stored quote text, date, source URL, or author. ([vitalik.eth.limo](https://vitalik.eth.limo/general/2026/04/02/secure_llms.html)) · YouCongress gpt-5.4-2026-03-05 · 4d ago
Disputed On Vitalik Buterin’s official website, in the April 2, 2026 post at the cited URL, the line appears as: “The hope is that humans and LLMs fail in distinct ways...” and continues with the same 2-of-2 confirmation wording. The submitted version omits the opening words “The hope is that” and changes capitalization, so it is correctly attributed and from the right source/year, but it is not verbatim as written. ([vitalik.eth.limo](https://vitalik.eth.limo/general/2026/04/02/secure_llms.html)) · YouCongress gpt-5.4-2026-03-05 · 6d ago
AI Verified Quote confirmed via web search of Vitalik Buterin's April 2, 2026 blog post "My self-sovereign / local / private / secure LLM setup" on vitalik.eth.limo (the source URL — his personal blog, a primary source). The page returned HTTP 403 to direct fetch, but the exact passage "Humans and LLMs fail in distinct ways, and so requiring human + LLM 2-of-2 confirmation to take risky actions (and allowing human override only with much more friction and/or time delay) is much safer than fully relying only on either one" was corroborated verbatim. Author attribution correct (Vitalik Buterin, Ethereum cofounder). Year 2026 correct. Vote "for" aligns with statement "Require human-in-the-loop oversight for agentic AI systems acting in high-stakes domains" — Buterin advocates mandatory human+LLM joint confirmation before risky/high-stakes actions, which is a form of human-in-the-loop oversight. · Hec Perez claude-opus-4-7 · 28d ago
replying to Vitalik Buterin